ViewVC Help
View File | Revision Log | Show Annotations | Download File
/cvs/AnyEvent/Changes
(Generate patch)

Comparing AnyEvent/Changes (file contents):
Revision 1.58 by root, Sat Apr 26 02:27:30 2008 UTC vs.
Revision 1.258 by root, Thu Jul 2 23:22:34 2009 UTC

1Revision history for Perl extension AnyEvent. 1Revision history for Perl extension AnyEvent.
2 2
3 - load AnyEvent::Handle only on demand in AnyEvent::DNS,
4 so AnyEvent::Socket users have smaller memory footprint
5 in the common case.
6 - add AnyEvent::Handle->push_shutdown method.
7 - add an additional error message parameter to AnyEvent::Handle's
8 on_error callback (for TLS, $! is still available).
9 - make AnyEvent::Handle more robust against early conenction
10 failures (during new), and return C<undef> in that case
11 from the constructor.
12
134.451 Fri Jul 3 00:28:58 CEST 2009
14 - do not clear rbuf when shutting down an AnyEvent::Handle
15 object - doing so breaks AnyEvent::HTTP.
16
174.45 Mon Jun 29 22:59:26 CEST 2009
18 - a write error could cause AnyEvent::Handle to create
19 an I/O watcher with an undefined $fh.
20 - special-case mapped ipv4 addresses in both
21 AnyEvent::Socket::format_address and parse_address,
22 to treat them just like ipv4 addresses.
23 - updated and overhauled the AnyEvent::Intro doc.
24 - implement AnyEvent::Socket::format_ipv4/ipv6.
25 - slightly speed up portable_pipe/socketpair.
26 - expand condvar begin/end documentation.
27
284.42 Fri Jun 26 08:32:18 CEST 2009
29 - preliminary and neccesarily incomplete support for IO::Async.
30 - reset SIGCHLD to DEFAULT when AnyEvent is loaded, in
31 case it was set to IGNORE, to ensure we can catch child statuses
32 even when the calling env acts stupidly.
33 - updated benchmarks with IO::Async, which performs very well.
34
354.412 Wed Jun 24 01:35:57 CEST 2009
36 - support an "untaint" attribute for AnyEvent::DNS
37 and set it on the default resolver.
38 - implement and document AnyEvent::Impl::Perl::loop.
39 - remove all anyevent-env variables from %ENV
40 when running in tainted mode.
41 - mention and extend the IO::Lambda benchmark.
42
434.411 Sun Jun 7 18:48:02 CEST 2009
44 - do not try to use F_SETFD on windows.
45
464.41 Thu May 14 06:40:11 CEST 2009
47 - work around issues in older perls (5.8.0?) when
48 a signal handler is deleted from the %SIG hash.
49 - use POSIX::_exit in child test, to avoid running
50 destructors.
51 - speed up CHLD handling by relying on SIGCHLD being
52 synchronously delivered, even when we roll our own
53 implementation.
54 - AnyEvent::DNS: add the "dname" resource record name
55 for cosmetic reasons.
56
574.4 Sun Apr 26 20:12:33 CEST 2009
58 - implemented idle watchers, where applicable.
59 - AnyEvent->time died when Event backend was in use.
60 - fix a memleak in the Tk backend.
61 - sped up Tk timer handling.
62 - clip negative "after" values to 0 in AnyEvent::Impl::Event
63 to avoid spamming out warning messages.
64 - fix Qt timers without interval.
65 - avoid Qt zero-timeout special case, in old libqt's.
66
674.352 Mon Apr 20 16:31:11 CEST 2009
68 - fix AnyEvent::Strict error messages for child watchers.
69 - fix/update Makefile.PL "recommends" versions.
70 - add AnyEvent->now_update.
71
724.351 Sat Apr 11 07:56:14 CEST 2009
73 - actually make the signal pipe work under win32.
74 - localise $! in signal handler to avoid changing $!, although
75 perl itself does not seem to save/restore errno either.
76 - set the cloexec flag on the signal pipe (normally set by perl too).
77
784.35 Fri Mar 27 11:48:20 CET 2009
79 - event models relying on AnyEvent's signal watcher emulation did
80 invoke the callback asynchronously, contradicting documentation
81 and causing signals to get lost (this includes AnyEvent's own
82 event loop). AnyEvent now uses the standard pipe trick to make
83 callback execution synchronous to the event loop.
84 - AnyEvent::Handle didn't free TLS context data on DESTROY
85 (patch by Pavel Shaydo).
86 - work around the ever-incompatibly-changing API of MakeMaker.
87 - document that changing global variables without restoring them
88 is a bad idea in Perl (noted by Adam Rosenstein).
89 - AnyEvent::Strict now barfs if ->io is passed a file.
90
914.34 Thu Feb 12 18:32:45 CET 2009
92 - separately buffer TLS read data, as otherwise the read queue could
93 deadlock as receiving data is not expected while draining the
94 read queue (which cna only happen with TLS).
95 - raise EBADMSG error on JSON decoding errors.
96 - fix some minor manpage bugs (reported by Maximilian Gaß).
97 - speed up select bitmask parsing quite a bit in the pure perl
98 backend.
99 - use CORE::select instead of just select in AnyEvent::Impl::Perl.
100
1014.331 Tue Jan 6 21:07:25 CET 2009
102 - socketpair fails on many vista machines because vista has
103 completely broken accept/getpeername and getsockname functions,
104 so we provide our own socketpair emulation that kind of works
105 (AnyEvent::Util::portable_pipe).
106 - new function: AnyEvent::Util::portable_socketpair.
107 - take advantage of the Guard module if it exists.
108
1094.33 Fri Nov 21 02:35:40 CET 2008
110 - AnyEvent::Strict did errornously flag a fileno of 0 as
111 illegal.
112 - reduce memory usage and slightly speed up the pure perl backend
113 by only storing the file descriptor, not the file handle.
114 - add missing autocork method to AnyEvent::Handle (reported
115 by Adam Rosenstein).
116 - AnyEvent::DNS->resolve errornously documented an $rcode result
117 argument, but there isn't (spotted by Henrik Krohns).
118 - the naked truth about Tk - it's basically unmaintained.
119
1204.32 Mon Nov 3 22:46:32 CET 2008
121 - fix AnyEvent::Socket::resolve_sockaddr to properly support
122 unix sockets again and choose a proper default for
123 the socket type.
124 - fix call to resolve_sockaddr in tcp_connect.
125
1264.31 Thu Oct 30 04:41:48 CET 2008
127 - implemented AnyEvent::Handle->destroy method.
128 - hint about unexpected effects in TLS mode.
129 - speed up AnyEvent::Impl::Perl by using more arrays
130 instead of hashes without the slightest loss of readability :)
131 - work around a perl argument refcounting bug.
132 - with some perl patches applied, AnyEvent now seems leak-free.
133
1344.3 Fri Oct 3 09:18:43 CEST 2008
135 - AnyEvent will now install a no-op signal handler for SIGPIPE,
136 unless one has been installed already.
137 - warn about not loaidng AnyEvent::Impl::POE early enough
138 (patch by Adam Rosenstein).
139 - fixed a great number of bugs and corner cases in AnyEvent::Handle:
140 - fix a bug in where in SSL connect mode, the
141 client would first wait for some data by the server and
142 otherwise hang. (reported and analyzed in an absolutely
143 exemplary manner by Adam Rosenstein).
144 - fix a bug in where SSL EOF would not be treated
145 as stream EOF, putting the connection into a hung state.
146 (reported and analyzed in an absolutely exemplary manner by Adam
147 Rosenstein).
148 - fix a potential segfault when the TLS context
149 would go missing in a read callback (e.g. due to stoptls) -
150 Net::SSLeay of course makes no type checking whatsoever.
151 - AnyEvent::Handle will not stop the read watcher in TLS mode.
152 - AnyEvent::Handle->stoptls will send a TLS close notify
153 instead of simply destroying the stream now.
154 - fix a bug where large blocks of data written in TLS mode
155 would not be sent unless triggered by receives.
156 - on_drain will now take the tls write buffer into account.
157 - SSL operations have been streamlined a bit and should be faster
158 now, more to come, though.
159 - added a FAQ section, will document nontrivial issues in there.
160
1614.234 Mon Sep 29 04:08:13 CEST 2008
162 - fix child watcher documentation: only child exits will be reported,
163 no trace events.
164 - mention SIGPIPE in AnyEvent::Handle.
165 - perl backend will now use POSIX::times as monotonic clock
166 when available and the monotonic clock option is not.
167 - run even without Time::HiRes in some cases now, but do not
168 provide sub-second accuracy (all *supported* perl versions have
169 Time::HiRes).
170
1714.233 Fri Aug 22 01:48:05 CEST 2008
172 - fix a bug in the testsuite, causing a hang.
173 - clarified lots of AnyEvent::Handle constructor arguments.
174
1754.232 Thu Aug 21 20:44:25 CEST 2008
176 - fix a bug in AnyEvent::Handle that could cause two fatal error
177 callbacks on EOF, which caused AnyEvent::HTTP to signal success twice
178 (testcase by Leon Brocard).
179 - clarified on_eof/on_error documentation a bit.
180
1814.231 Tue Jul 29 13:12:15 CEST 2008
182 - remove some debugging code left in AnyEvent::Util::fork_call
183 (and no, it's impossible to implement with the broken windows
184 perls without resource leaks or worse).
185
1864.23 Tue Jul 29 12:19:59 CEST 2008
187 - document the first parameter passed to condvar callbacks to be
188 the callback.
189 - add AnyEvent::Socket::{ntoa,aton} aliases.
190 - optimize the AE::Handle->push_read (line) for the default
191 eol marker.
192 - optimize push_read (packstring|storable) for small packets.
193 - invoke on_error callback when no on_eof callback is set.
194 - fix a bug in push_read (storable) of unknown impact.
195
1964.22 Sun Jul 20 16:34:13 CEST 2008
197 - new function AnyEvent::Socket::parse_hostport.
198 - as the bulkheads at microsoft can't even get getprotobyname reliably
199 working on their shitty fucking broken os we need to hardcode
200 some common protocol numbers in AnyEvent::Socket. How can
201 people even bother with such a pile of shit as windows.
202
2034.21 Thu Jul 17 14:40:05 CEST 2008
204 - INCOMPATIBLE CHANGE: AnyEvent::Util::fork_call now has a prototype
205 which is incompatible to previous invocation syntax.
206 - work around more windows perl bugs in fork_call. I cannot imagine
207 why anybody would bother implementing fork in such an obviously
208 unusable way.
209 - avoid calling $do_retry if the request has already finished in
210 AnyEvent::DNS, thus avoiding an exception (reported by Anatoly
211 K. Sharifulin).
212
2134.2 Sat Jul 12 22:42:11 CEST 2008
214 - support an "interval" argument for repeating timers.
215 - fix a bug in the pure perl event loop that caused timers to
216 be delayed under some circumstances.
217 - implement optional argument checking via PERL_ANYEVENT_STRICT.
218 - not importing anything from Event causes it to malfunction
219 and not use Time::HiRes, work around this bug.
220 - more bug workarounds against the endlessly broken Tk module,
221 seems to work now on 32 bit machines, but less so on 64 bit.
222 - minor optimisations applied to most backends.
223
2244.161 Fri Jul 4 14:24:48 CEST 2008
225 - fixed the t/handle/02_write.t, it read after write, with the
226 assumption that all data could be written before read
227 (reported by Jim Razmus).
228
2294.160 Thu Jul 3 04:02:21 CEST 2008
230 - re-registering signal watchers after unregistering one did
231 not work if a backend falls back on AnyEvent's default
232 implementation (also affected child watchers).
233 - new AnyEvent::Handle options: autocork and no_delay.
234
2354.152 Sun Jun 22 14:15:44 CEST 2008
236 - allow for 32-bit perls that implement shifts differently
237 on different architectures in parse_ipv4 (reported and
238 analysed by Keiichi DAIBA).
239
2404.151 Fri Jun 6 17:34:24 CEST 2008
241 - make sure specifying _only_ on_read and never pushing reads
242 works.
243
2444.15 Fri Jun 6 13:00:46 CEST 2008
245 - the pure perl backend would keep some watchers alive when more than
246 one watcher was registered for the same fd.
247 - new "packstring" and "storable" read and write types
248 for AnyEvent::Handle.
249 - allow on_eof handler to be called after on_error with EPIPE returns.
250 - do not immediately call on_read callback in handle constructor.
251
2524.14 Thu Jun 5 20:29:31 CEST 2008
253 - Fixed a bug in DNS SRV priority sorting.
254 - AnyEvent::Util::guard now reports runtime errors while
255 executing the guard block as warnings.
256 - handle 0-byte-reads just before EOF correctly in AnyEvent::Handle.
257
2584.13 Thu Jun 5 00:47:59 CEST 2008
259 - AnyEvent::DNS only followed cname chains with length 2,
260 contrary to documentation. bumped it up to 10, thanks to
261 microsoft, the current broken-dns-config-king.
262 - AnyEvent::DNS didn't check for socket return status
263 but instead relied on perl not creating filehandles in that
264 case - too bad it gives you a bogus file handle (reported
265 and analysed by Vladimir Timofeev).
266 - fix queue management logic in AnyEvent::Handle:
267 when on_read was registered, the queue was empty
268 and no progress could be made AnyEvent::Handle would
269 enter an endless loop.
270 - correctly start reading again when the handle became
271 busy again after idling.
272 - correctly treat tls shut-downs as EOF condition
273 (for the time being, would be nice to have a callback for that).
274 - correctly call eof callback under all conditions
275 (hopefully).
276 - the timeout callback did not expect that $self can go
277 away any time.
278
2794.12 Tue Jun 3 10:58:04 CEST 2008
280 - include AnyEvent::Intro, a tutorial for anyevent,
281 anyevent::socket and anyevent::handle.
282 - allow more options in on_error.
283
2844.11 Fri May 30 23:42:25 CEST 2008
285 - INCOMPATIBLE CHANGE: replace ptr by real PTR lookup, provide
286 reverse_lookup and reverse_verify to replace it, support
287 v4mapped and v4compat addresses.
288 - provide more documentation for the resolver class.
289 - really replace longest run of :0: by :: in format_address,
290 also properly convert :: and ::1 again.
291 - support NAPTR record name and decode it.
292 - implement random weight sampling for SRV records, as per
293 rfc 2782.
294 - correctly abort on srv-record targets of ".".
295 - added AnyEvent::DNS::wait_for_slot.
296 - in the unlikely event of a virtual circuit connection
297 being invalidated by a delayed udp reply, AnyEvent::DNS
298 could die. this has been fixed.
299 - plug a probable memleak in the DNS vc code.
300 - use configured timeout also for VC request phase.
301 - implement timeout and max_outstanding methods
302 for resolver class.
303 - update version numbers in all modules.
304
3054.1 Thu May 29 05:45:40 CEST 2008
306 - INCOMPATIBLE CHANGE: renamed xxx_ip functions to xxx_address
307 which mirrors their purpose better (old names still available
308 for a while). Also moved AnyEvent::DNS::addr to
309 AnyEvent::Socket::resolve_sockaddr.
310 - implement AnyEvent->time and AnyEvent->now.
311 - fix IPv6 support in pack_sockaddr.
312 - officially un-experimentalise all newly introduced code.
313 - support unix domain sockets everywhere by specifying
314 a host of "unix/" and the pathname as service.
315 - implement an activity timeout in AnyEvent::Handle.
316 - added a regex read type to AnyEvent::Handle.
317 - added a json read/write type to AnyEvent::Handle.
318 - always croak in on_error in AnyEvent::Handle.
319 - document how to subclass AnyEvent::Handle.
320 - implement AnyEvent::Util::fork_call.
321 - add support for IPv6 nameservers and nameserver statements.
322 - work around _yet_ _another_ windows perl bug in where empty
323 select masks cause errors in the pure perl backend (this
324 workaround cannot be implemented for the other event loops,
325 but EV also works around this).
326 - supply AnyEvent::Socket::address_type and make good use of it.
327 - clarify and fix the Handle documentation w.r.t $self vs. $handle
328 vs. callback arguments.
329 - add some recommends to the META.yml.
330
3314.05 Mon May 26 19:44:06 CEST 2008
332 - some platforms ignore the rfc and prepend an extra sa_len member to
333 the sockaddr structure, cater for those. also use sockaddr_family,
334 didn't know it was there, but comes in handy for the workaround.
335 - undo the import WIN32 hack.
336
3374.04 Mon May 26 08:03:31 CEST 2008
338 - try to work around yet another windows bug: failed connects are reported
339 as if it were out-of-band data. windows users: you suck. Only
340 EV, Glib, Event and the pure perl backend can handle this condition.
341 - optimize write algorithm in AnyEvent::Handle.
342 - properly parse PERL_ANYEVENT_PROTOCOLS and use the info correctly.
343 - tcp_connect now properly iterates through all targets.
344 - check wether IPv6 sockets can actually be created, otherwise, disable ipv6
345 support entirely. also hardcode some AF_INET6 constants for the
346 sake of perl 5.8.
347 - vastly improved nameserver/suffix detection algorithm for windows
348 (still a hack...)
349 - try to use a fallback dns server if no dns servers could be found.
350 - splatter around some local $SIG{__DIE__}'s to work around bad
351 die handlers.
352
3534.03 Sun May 25 05:05:57 CEST 2008
354 - work around perl bugs on windows, where perl returns undocumented
355 error codes for sysread, syswrite etc.
356 - fix AnyEvent::Handle::unshift_read to really unshift and not push.
357 - AnyEvent::Handle could sometimes stumble over 0-byte writes.
358 - fix unpack_sockaddr, now works for AF_INET6 sockaddr's.
359 - updated documentation.
360 - lots of workarounds for perl 5.8 quirks.
361 - implement a more flexible type system for both reads and writes
362 in AnyEvent::Handle.
363 - support IPv6 in tcp_server.
364 - pass local hostname and port to the prepare callback in tcp_server.
365 - make types extendable in AnyEvent::Handle.
366 - croak, not die, in AnyEvent::Handle;
367
3684.0 Sat May 24 19:58:08 CEST 2008
369 - added (experimental) AnyEvent::DNS module.
370 - added (experimental) AnyEvent::Socket module.
371 - added (experiemntal) support for TLS/SSL in AnyEvent::Handle.
372 - make condvar->end call ->send as documented.
373 - make condvar constructor support the documented cb => argument.
374 - overload condvars so you can use them instead of code references.
375 - support more of the POSIX forms instead of a pure dotted quad for
376 verbatim ip addresses.
377 - AnyEvent::Socket::inet_aton will use AnyEvent::DNS now and not fork anymore.
378 - try very hard not to freeze on broken windows perls.
379 - make on_eof optional in AnyEvent::Handle.
380 - support service names in tcp_connect and tcp_server.
381 - work around netbsd/openbsd bug causing unix sockets not to
382 become writable until empty.
383 - only make a dummy read for error detection when getpeername
384 returns ENOTCONN.
385 - defend AnyEvent::Util::inet_aton against Coro::LWP's brutal
386 override.
387 - new AnyEvent::Guard::cancel.
388
3893.5 Sat May 17 23:17:14 CEST 2008
390 - gracefully deal with systems not defining CLOCK_MONOTONIC
391 (as opposed to not supporting the clock itself, which worked in 3.41).
392 - added AnyEvent::Util::tcp_connect and AnyEvent::Util::tcp_server
393 helpers, to replace the ill-designed AnyEvent::Socket class.
394 - added some examples (eg/connect, eg/handle).
395 - added AnyEvent::Util::guard.
396 - AnyEvent::Util now exports most of its functions.
397 - fixed a bug in *_read_chunk, where the callback was called with $self as
398 first and second argument.
399 - simplified and fixed regex handling in AnyEvent::Handle::*_read_line.
400 - fix manpage condvars examples to use new syntax.
401 - considerably improve inet_aton with EV::ADNS backend to support
402 cname chains (common dns configuration bug, e.g. with www.google.de)
403 and aaaa records if no a record could be found.
404
4053.41 Sun May 11 19:53:13 CEST 2008
406 - work around a bug in perl 5.8.8's local. Reported by Yi Ma Mao.
407
4083.4 Sun May 11 00:29:25 CEST 2008
409 - complete re-implementation of condvars with a hopefully much more useful
410 API (backward compatible functions exist but are undocumented).
411 - AnyEvent::Handle fully rewritten with a hopefully more powerful API
412 (still subject to change).
413 - take advantage of CLOCK_MONOTONIC, if available, in AnyEvent::Impl::Perl.
414 - provide AnyEvent::post_detect and @AnyEvent::post_detect, which
415 allows module authors to avoid forcing event loop detection.
416 (used by Coro::AIO, Coro::BDB, Coro::AnyEvent for example).
417 - remove coro backends: Coro now provides generic support for AnyEvent
418 via Coro::AnyEvent.
419
4203.3 Mon Apr 28 09:51:06 CEST 2008
421 - added AnyEvent::Handle, AnyEvent::Socket and AnyEvent::Util modules.
3 - fix a bug in the pure perl backend that kept watchers alive 422 - fix a bug in the pure perl backend that kept watchers alive
4 when multiple watchers were registered for the same fh/poll combo. 423 when multiple watchers were registered for the same fh/poll combo.
5 - add a benchmark section showing AnyEvent overhead and comparing 424 - add a benchmark section showing AnyEvent overhead and comparing
6 the different event loops with each other. 425 the different event loops with each other.
7 - prefer pure perl over tk when autoprobing, as it's about as fast, 426 - prefer pure perl over tk when autoprobing, as it's about as fast,
8 but doesn't crash with many watchers. 427 but doesn't crash with many watchers.
9 - make Qt support non-experimental. 428 - declare Qt support non-experimental.
10 - clarify the confusing section about the file descriptor being 429 - clarify the confusing section about the file descriptor being
11 kept alive. 430 kept alive.
12 - document the race between loading of an event module and 431 - document the race between loading of an event module and
13 child processes exiting. 432 child processes exiting.
14 - support POE as "backend" (with some caveats, POE is not generic 433 - support POE as "backend" (with some caveats, POE is not generic
15 enough, and darn slow). 434 enough, and darn slow).
16 - support Wx and Prima through POE. 435 - support Wx and Prima through POE.
17 - optimise perl backend to use 20% less memory and to be roughly 436 - optimise perl backend to use 20% less memory and take advantage
18 2-3 times faster in the common case of sparsely-active file 437 of typical timeout behaviour. It can now compete with
19 descriptors. It can now compete with select/poll-based C event 438 select/poll-based C event loops in most cases (it is usually
20 loops in most cases (it is usually faster than Event and Glib :). 439 faster than Event and Glib :).
21 - roughly cut EV memory use in half and icrease its speed by 30%, 440 - roughly cut EV memory use in half and increase its speed by 30%,
22 by removing undocumented functionality. 441 by removing undocumented functionality. Did something similar
442 to other event loops where possible.
23 443
243.2 Thu Apr 24 10:10:40 CEST 2008 4443.2 Thu Apr 24 10:10:40 CEST 2008
25 - do not die when anyevent watchers are destroyed while running callbacks 445 - do not die when anyevent watchers are destroyed while running callbacks
26 in the pureperl backend (could only happen when two watchers are registered 446 in the pureperl backend (could only happen when two watchers are registered
27 for the same fh/poll combo). 447 for the same fh/poll combo).

Diff Legend

– Removed lines
+ Added lines
< Changed lines
> Changed lines